27 Nov
1992
27 Nov
'92
5:44 a.m.
In asking my previous question about comonads vs. adjunctions in linear logic I had assumed a negative answer to the following question: Need the Kleisli category be the only CCC among the adjunctions defining a Girard comonad !:D->D (one with !(AxB)^!(A)@!(B), !(1)^I)? What is the truth of the matter? Under the proposed definition of ! (just a comonad, no specification of the intended adjunction) a counterexample to the above would be ruled out as a model of linear logic in the sense that abstracting the adjunction to a comonad in effect replaces the intended CCC by the (distinct) Kleisli CCC. Vaughan Pratt ==============================================================================